Urban underground safety risk assessment method and device based on dynamic updating algorithm, electronic equipment and storage medium

Abstract

The invention relates to the field of urban underground safety management, and discloses an urban underground safety risk assessment method and device based on a dynamic updating algorithm, electronic equipment and a storage medium, and the method comprises the steps: constructing an urban underground safety risk assessment matrix, and carrying out the quantification of risk factors; optimizing the weight of the risk factor, and dynamically adjusting the weight; constructing an improved Bayesian network model; reasoning by using a Bayesian network model, predicting the overall risk state of the underground system and identifying key risk factors; constructing a risk factor complex network model, and determining key risk factors; performing time sequence evolution modeling on the state of the risk factor, and dynamically updating the state of the risk factor in combination with real-time monitoring data; and finally, outputting an evaluation result which comprises an overall risk state, a key risk factor and a risk change trend of the underground system. According to the method, scientific quantification, real-time updating and dynamic evaluation of urban underground safety risks can be realized, and efficient and accurate technical support is provided for safety management of an underground system.

Description

Technical Field

[0001] The present invention relates to the technical field of urban underground safety management, and particularly to an urban underground safety risk assessment method, device, electronic device and storage medium based on a dynamic update algorithm. Background Art

[0002] With the acceleration of the urbanization process, the development and utilization of underground space have increased day by day. The construction of large-scale infrastructure such as subways, tunnels, and underground parking lots has significantly improved urban functions and land use efficiency. However, the complexity and dynamics of the urban underground system have also led to a substantial increase in potential safety risks. Problems such as foundation settlement, groundwater leakage, structural deformation, and human construction disturbances may cause local or systemic safety hazards. The effective assessment of these safety risks has become a key link in the design, construction, and operation management of underground projects.

[0003] In the prior art, most of the assessment methods for urban underground safety risks rely on the experience of risk assessment personnel and lack quantitative analysis of risk factors. This method is highly subjective, and the assessment results are easily affected by the judgment criteria of the assessment personnel, making it difficult to ensure objectivity and consistency. In addition, traditional assessment methods often cannot comprehensively analyze the coupling effect and causal relationship of multiple risk factors in the underground system. Especially when facing dynamic environmental changes, they lack corresponding dynamic modeling and real-time adjustment capabilities, resulting in a deviation between the assessment results and the actual situation. In addition, existing methods usually do not fully combine real-time monitoring data and complex network analysis technology, making it difficult to accurately identify the risk factors that play a key role in system safety, thus affecting the effectiveness of risk management decisions.

[0004] Based on the above problems, there is an urgent need in the prior art for a scientific assessment method that can dynamically adapt to complex environmental changes, combine the causal relationship and multi-dimensional characteristics of risk factors, and support real-time data update, so as to achieve a more accurate and reliable urban underground safety risk assessment. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides an urban underground safety risk assessment method, device, electronic device and storage medium based on a dynamic update algorithm, and solves the technical problems of strong subjectivity, poor dynamic adaptability, and inability to accurately identify key risk factors in the prior art.

[0006] To achieve the above object, the present invention is realized through the following technical solutions: An urban underground safety risk assessment method based on a dynamic update algorithm includes the following steps: Construct an urban underground safety risk assessment matrix, and quantify the risk factors of urban underground safety based on the consequence severity level and occurrence probability level of the risk factors; Optimize the weights of risk factors. Using the entropy value theory, dynamically adjust the weights according to the contribution of risk factors to the evaluation results; Based on risk factors and their causal relationships, construct an improved Bayesian network model, where risk factors are used as network nodes and causal relationships are used as network edges, and optimize the network structure by combining a dynamic causal relationship adjustment algorithm; Use the improved Bayesian network model for forward reasoning and backward reasoning to predict the safety risk status of the underground system and identify key risk factors respectively; Construct a complex network model of risk factors based on causal relationships, analyze the importance of network nodes, and determine key risk factors; Model the time series evolution of risk factors based on the theory of stochastic dynamic systems, and perform real-time update and adjustment on the dynamic changes of risk factors; Output the evaluation results, including the overall risk status of the underground system, key risk factors, and risk change trends.

[0007] Preferably, the steps of constructing the urban underground safety risk assessment matrix include: By dividing the consequence severity level of risk factors into multiple levels and the occurrence probability level into multiple levels, generate a two-dimensional matrix, where the comprehensive score of risk factors is the product of the consequence severity level and the occurrence probability level.

[0008] Preferably, the steps of optimizing the weights of risk factors include: Based on the entropy value theory, calculate the conditional entropy of each risk factor and the evaluation result, and normalize and adjust the weights of risk factors according to the conditional entropy value, so as to dynamically optimize the influence weights of risk factors.

[0009] Preferably, the steps of constructing an improved Bayesian network model based on risk factors and their causal relationships include: Through the Markov chain Monte Carlo algorithm, iteratively adjust the edge weights of the causal relationships between risk factors in the network, optimize the edge weights of the causal relationships, and construct a dynamic Bayesian network model based on the update of the causal relationships.

[0010] Preferably, the steps of using the improved Bayesian network model for forward reasoning and backward reasoning include: Based on the status of current risk factors, calculate the occurrence probability of potential risk consequences of the underground system; Based on the risk consequences of the underground system, reverse deduce the main risk factors that lead to the risk.

[0011] Preferably, the steps of constructing a complex network model of risk factors based on causal relationships include: Regarding risk factors as network nodes and the causal relationships between risk factors as network edges, a weighted scale-free network is constructed according to the weights of the causal relationships, and key risk factors are identified by analyzing the degree centrality and betweenness centrality of network nodes.

[0012] Preferably, the step of modeling the temporal evolution of risk factors based on the theory of stochastic dynamic systems and performing real-time update and adjustment on the dynamic changes of risk factors includes: By establishing a dynamic evolution model of the risk factor state, describing the change trend of risk factors, and combining real-time data to adjust the state probability of nodes and edge weights in the Bayesian network, the real-time update of the evaluation model is realized.

[0013] The present invention also provides an urban underground safety risk assessment device based on a dynamic update algorithm, including: A matrix construction module for quantifying the consequence severity level and occurrence probability level of risk factors and constructing a risk assessment matrix; A weight optimization module for dynamically adjusting the weights of risk factors based on the entropy value theory; A network construction module for constructing an improved Bayesian network based on risk factors and their causal relationships; An inference analysis module for performing forward inference and backward inference based on the Bayesian network model; A network analysis module for constructing a complex network based on causal relationships and identifying key risk factors; A dynamic update module for modeling and real-time updating the temporal evolution of risk factors.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method as described above is implemented.

[0015] The present invention also provides a storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method as described above is implemented.

[0016] The present invention provides an urban underground safety risk assessment method, device, electronic device, and storage medium based on a dynamic update algorithm. It has the following beneficial effects: 1. By constructing a risk assessment matrix and optimizing weights, the present invention quantifies the influence of risk factors based on the entropy value theory, eliminates the subjectivity of relying on expert experience in traditional assessment methods, makes the assessment results more scientific and objective, and provides a solid theoretical basis for the risk management of complex urban underground environments.

[0017] 2. The present invention models the causal relationships of risk factors through an improved Bayesian network and supports forward and backward reasoning analysis. It can not only predict the risk status but also trace and identify key risk factors, providing accurate path analysis and decision-making basis for risk control.

[0018] 3. The present invention introduces a complex network model. By analyzing the centrality index of nodes, it quantifies the importance and role of key risk factors in the whole network, and can accurately identify the factors that have the greatest impact on the overall safety of the system, providing a scientific basis for priority resource allocation and key control.

[0019] 4. The present invention models the state changes of risk factors based on the theory of stochastic dynamic systems and dynamically adjusts the model parameters in combination with real-time monitoring data to achieve real-time update and dynamic prediction of underground risks, enabling the evaluation system to adapt to the dynamic changes of complex environments and improving the accuracy and timeliness of the results.

[0020] 5. The present invention outputs the overall risk status, key risk factors, and risk change trends through comprehensive analysis of the evaluation results, and presents the evaluation results through visualization tools, helping users quickly understand and judge the risk status of the underground system, and providing intuitive and effective support for risk management and decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a schematic flow chart of the method of the present invention; Figure 2 is a schematic structural diagram of the device of the present invention; Figure 3 is a schematic structural diagram of the electronic device of the present invention.

[0022] Among them, 100 is a matrix construction module; 200 is a weight optimization module; 300 is a network construction module; 400 is an inference analysis module; 500 is a network analysis module; 600 is a dynamic update module; 40 is a computer device; 41 is a processor; 42 is a memory; 43 is a storage medium. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0024] Please refer to the attached Figure 1, the present invention provides a method for urban underground safety risk assessment based on a dynamic update algorithm. The purpose of the present invention is to achieve scientific, accurate, and real-time assessment of urban underground safety by constructing a quantification model and a dynamic update algorithm.

[0025] As Figure 1 shown, the method for urban underground safety risk assessment may include the following steps: S1. Construct an urban underground safety risk assessment matrix; S2. Optimize the weights of risk factors; S3. Construct an improved Bayesian network model; S4. Conduct inference analysis using the improved Bayesian network model; S5. Construct a complex network model and analyze key factors; S6. Based on the time series evolution and dynamic update of a stochastic dynamic system; S7. Output the assessment results.

[0026] The following details each step of the method for urban underground safety risk assessment based on the dynamic update algorithm.

[0027] For step S1, in this embodiment, constructing an urban underground safety risk assessment matrix aims to provide a standardized and data-based foundation for the input of subsequent risk assessment models through the quantification of the consequence severity level and occurrence probability level of risk factors.

[0028] It should be noted that the risk assessment matrix takes the consequence severity and occurrence probability of risk factors as the main dimensions to form a two-dimensional matrix, thereby realizing the quantitative expression of risk factors. Each element in the matrix represents the comprehensive risk level of a certain risk factor under specific conditions.

[0029] In some embodiments, the consequence severity level of risk factors can be classified according to specific urban underground safety scenarios. For example, typical risk factors in the underground system include but are not limited to: Geological conditions, such as the stability of rock formations and the bearing capacity of the foundation; Structural status, such as the deformation of tunnel linings and the width of cracks; Hydrological environment, such as the seepage velocity of groundwater and the water level height; Human activities, such as vibrations and excavation disturbances during construction.

[0030] Exemplarily, the consequence severity level can be divided into three levels: C 1 : Slight impact, for example, rock cracks do not expand and do not affect structural safety; C 2: Medium impact, for example, groundwater seepage causes certain erosion to the structure but does not lead to structural failure; C 3 : Severe impact, for example, foundation settlement causes structural instability.

[0031] As an option, the probability of occurrence level can be divided into the following levels: P 1 : Low probability, usually less than 5%; P 2 : Medium probability, usually between 5% and 20%; P 3 : High probability, usually greater than 20%.

[0032] It should be noted that the specific severity level and probability of occurrence level can be adjusted according to the actual situation of different underground systems.

[0033] In this embodiment, the construction of the risk assessment matrix includes combining the consequence severity level corresponding to each risk factor with the probability of occurrence level to form matrix elements.

[0034] Specifically, each element R in the matrix ij represents the comprehensive score of a certain risk factor. As a possible implementation, the comprehensive score can be calculated by the weighted product of the consequence severity level and the probability of occurrence level: R ij = C i ·P j Where: C i represents the consequence severity level of the risk factor; P j represents the probability of occurrence level of the risk factor.

[0035] In some embodiments, the above formula can also be extended. For example, when the influence of some risk factors needs to be further weighed, a weight coefficient W k can be added to correct the result of the comprehensive score: R ij = W k ·C i ·P j It should be noted that the value of the weight coefficient W k is determined according to the importance of the risk factor to the overall evaluation result, and the specific weight assignment will be optimized in the subsequent steps.

[0036] In a possible implementation, the construction of the evaluation matrix can be achieved by the following method: First, based on the historical data of the urban underground system, the consequence severity and occurrence probability of each risk factor are statistically analyzed, classified, and graded. Secondly, the comprehensive score of each risk factor is calculated using the grading rules, and the corresponding positions are filled in the two-dimensional matrix.

[0037] Exemplarily, for a specific underground risk factor (such as the change in the groundwater level), its severity level is C 2 , and the occurrence probability level is P 3 . Then, the comprehensive score of this factor is: R ij = C 2 ·P 3 Assume C 2 = 2, P 3 = 3. Then the comprehensive score R ij = 6. In the matrix, the corresponding position value is filled with 6.

[0038] It can be understood that the method for constructing the risk assessment matrix in this embodiment has strong flexibility, and users can adjust the classification rules, scoring methods, and comprehensive calculation formulas of the matrix according to actual needs.

[0039] It should be noted that the construction result of the matrix not only quantitatively expresses the risk factors but also provides a data basis for subsequent dynamic optimization and network modeling. Therefore, the accuracy of the matrix directly affects the accuracy of the assessment results.

[0040] As an option, the risk assessment matrix can be implemented through an automated tool. For example, a data processing program can extract risk factor data from sensor data, historical records, and expert evaluations, and automatically complete matrix filling and scoring calculations.

[0041] In some embodiments, the results of the risk assessment matrix can be further visualized. For example, a two-dimensional chart is constructed, and the consequence severity and occurrence probability of the risk factors are represented by matrix cells of different colors or sizes. Exemplarily, high-score cells can be marked in red to indicate high-risk areas.

[0042] For step S2, in this embodiment, by optimizing the weights of the risk factors, the accuracy and scientificity of the urban underground safety risk assessment are further improved. Specifically, in this embodiment, the entropy value theory is used to dynamically adjust the weights of the risk factors, so as to accurately reflect the actual contribution degree of each risk factor to the assessment results.

[0043] It should be noted that the core objective of weight optimization is to reduce the influence of human subjectivity, so that the weight distribution can change dynamically based on the correlation between the risk factors and the assessment results, and at the same time provide more reliable input data for the subsequent steps.

[0044] In some embodiments, the initial step of weight optimization is to conduct a quantitative analysis of the correlation between risk factors and evaluation results. To this end, the concept of conditional entropy is introduced in this embodiment, and the dependence relationship between each risk factor and the overall risk status is quantified by calculating the conditional entropy of each risk factor with respect to the evaluation result.

[0045] Exemplarily, the calculation formula of the conditional entropy H(X i |Y) is as follows: Where: X i represents a certain risk factor; Y k represents the possible states of the evaluation result; P(X i ,Y k ) represents the joint probability of the risk factor X i and the evaluation result Y k .

[0046] It should be noted that the smaller the conditional entropy value, the greater the contribution of the risk factor X i to the evaluation result Y. Therefore, the conditional entropy can effectively quantify the influence degree of the risk factor and provide a scientific basis for subsequent weight allocation.

[0047] In this embodiment, the weight allocation based on conditional entropy is achieved through the following formula: Where: w i represents the weight of the risk factor X i ; H(X i |Y) represents the conditional entropy of the risk factor X i ; n represents the total number of risk factors.

[0048] As an option, the normalization method in the weight allocation formula can be adjusted according to actual needs. For example, by adding an additional weight correction factor α i , so that the weight allocation is more adaptable to specific scenarios. The adjusted formula is: It should be noted that the value of the correction factor α i can be dynamically set according to expert experience or actual data.

[0049] In a possible implementation, the weight optimization process in this embodiment can be dynamically updated by combining historical data and real-time data. For example, during actual operation, when the state of a certain risk factor in the real-time monitoring data changes significantly, the conditional entropy value can be recalculated and the weight can be updated to ensure the real-time nature of the evaluation.

[0050] It can be understood that the dynamic update mechanism enables the weight optimization to adapt to the complexity and variability of the underground safety system, thereby improving the applicability of the evaluation model.

[0051] It should be noted that the optimized weights are not only used to adjust the score calculation in the risk assessment matrix, but also serve as input parameters for the subsequent construction of the Bayesian network to ensure the accuracy of the causal relationship in the Bayesian network.

[0052] In some embodiments, the results of the weight optimization can be visually displayed. For example, the weight allocation results of the risk factors are presented in the form of a bar chart or a pie chart, which is convenient for users to intuitively understand the relative importance of each risk factor.

[0053] The weight optimization method of this embodiment can effectively solve the problem of fixed and inflexible weight allocation in traditional evaluation methods. Through data-driven dynamic adjustment, the risk assessment results are made more reliable.

[0054] For step S3, in this embodiment, by constructing an improved Bayesian network model, it aims to model the causal relationship between risk factors and realize the dynamic analysis and optimization of risk assessment. Specifically, the Bayesian network model represents risk factors as nodes and causal relationships as edges through a mathematical graph structure, forming a quantifiable causal relationship model.

[0055] It should be noted that the construction of the Bayesian network model includes two parts: the definition of the topological structure and the optimization of probability parameters, which combines the static association and dynamic adjustment mechanism between risk factors. This embodiment uses the Markov Chain Monte Carlo (MCMC) algorithm to optimize the network, thereby improving the accuracy and adaptability of the model.

[0056] In some embodiments, the topological structure of the initial Bayesian network can be determined according to the actual situation of the urban underground safety scenario. For example, certain risk factors (such as changes in the groundwater level) may directly affect other risk factors (such as foundation settlement). Specifically, the steps for constructing the initial Bayesian network include the following: Determine the node set V of the network, where each node represents a risk factor, such as foundation bearing capacity, changes in the groundwater level, or structural deformation.

[0057] Determine the causal relationships between nodes to form the edge set E. Exemplarily, the change in the groundwater level may affect the structural deformation through seepage effects, and further affect the overall risk state.

[0058] It should be noted that the setting of the initial topological structure can be based on expert knowledge, historical data, or existing theoretical models, serving as the basis for subsequent optimization.

[0059] In this embodiment, the joint probability distribution of the Bayesian network is used to describe the probability relationships of all nodes (i.e., risk factors) in the network. The definition of the joint probability distribution is as follows: Where: X i represents a certain risk factor; Pa(X i ) represents the set of parent nodes of node X i (i.e., the factors that directly affect X i ); P(X i |Pa(X i )) represents the conditional probability of node X i given the states of its parent nodes.

[0060] It should be noted that this formula ensures that the joint distribution of risk factors can be decomposed through conditional independence, thus simplifying the complexity of multi-dimensional risk factor modeling.

[0061] To further optimize the Bayesian network model, the Markov Chain Monte Carlo (MCMC) algorithm is used in this embodiment to dynamically adjust the topological structure and conditional probability parameters of the network. In one possible implementation, the optimization process includes the following: According to the data input of the risk assessment matrix, the edge weight w ij is iteratively updated. The specific update formula is as follows: Where: represents the weight of edge (i,j) in the t-th iteration; η is the learning rate, used to control the optimization step size; is the gradient of the joint probability distribution with respect to the edge weight.

[0062] Through iterative adjustment, the network can gradually approach the optimal structure, making the causal relationships of risk factors in the network more in line with reality.

[0063] It should be noted that this dynamic optimization method can handle the changes in the relationships of risk factors under different underground scenarios, improving the applicability of the model.

[0064] As an option, a regularization term can also be introduced in this embodiment to avoid the overfitting problem in the Bayesian network model. For example, for the redundant edges that may exist in the network, the following formula can be used for constraint optimization: Where: L is the objective function, including a data fitting term and a regularization term; λ is the regularization coefficient, used to control the model complexity; |w ij | is the absolute value of the edge weight.

[0065] It should be noted that the improved Bayesian network in this embodiment can not only statically model the causal relationship between risk factors, but also adapt to the changes of real-time monitoring data through a dynamic optimization mechanism, so as to more accurately reflect the risk state of the underground system.

[0066] In some embodiments, the constructed Bayesian network model can be displayed through a visualization tool. For example, the graphical interface is used to display the network topology structure, where the nodes represent risk factors, and the thickness of the edges represents the strength of the causal relationship.

[0067] It can be understood that the method for constructing the Bayesian network model in this embodiment can provide basic support for subsequent inference and analysis.

[0068] For step S4, in this embodiment, the improved Bayesian network model constructed above is used to analyze the urban underground safety risk through two methods: forward reasoning and backward reasoning, so as to predict the safety risk state of the underground system and identify key risk factors respectively. It should be noted that forward reasoning and backward reasoning are the core application functions of the Bayesian network model, which can associate the state of risk factors with the overall evaluation result, and thus provide a scientific basis for risk management and control.

[0069] In some embodiments, forward reasoning is used to predict the overall safety risk state of the system based on the known state of risk factors. Specifically, the core goal of forward reasoning is to calculate the occurrence probability P(Y) of the evaluation result Y, and its calculation is based on the joint probability distribution formula of the Bayesian network: Where: X i represents the i-th risk factor; P(X i ) represents the probability of the risk factor X i ; P(Y|X 1 ,X 2,...,X n ) represents the conditional probability of the evaluation result Y given that the states of all risk factors are known.

[0070] As a possible implementation, under the influence of multiple factors such as groundwater level changes and foundation settlement, forward reasoning can quantify the overall safety state. For example, when the probability distribution of the groundwater level is known and the state of foundation settlement is updated in real time through sensor data, forward reasoning can dynamically calculate the probability that the underground system is in a "high-risk" or "low-risk" state.

[0071] It should be noted that the results of forward reasoning can be presented in a visual form, such as a probability curve graph or a heat map of the risk state, so as to help users quickly understand the risk level of the system.

[0072] In this embodiment, backward reasoning is used to reverse-derive the main risk factors that lead to the known risk assessment result from the known risk assessment result. Specifically, the core goal of backward reasoning is to calculate the conditional probability P(X i |Y) of each risk factor X given the evaluation result Y. Its calculation formula is: i |Y). Where: P(X i |Y) represents the conditional probability of the evaluation result Y given the state of the risk factor X i ; P(X i ) represents the prior probability of the risk factor X i ; P(Y) is the marginal probability of the evaluation result Y, which can be calculated through forward reasoning.

[0073] In some embodiments, the implementation of backward reasoning can be combined with the identification of key nodes. For example, when the evaluation result is "high risk", backward reasoning is used to calculate the probability of each risk factor leading to high risk and sort them according to the probability size, so as to identify the factor that has the greatest impact on the overall risk.

[0074] As an option, the calculations of forward reasoning and backward reasoning can be implemented in a discretized or continuous manner. For example: In the discretized implementation, the states of the risk factors are divided into a finite number of levels (such as low, medium, high), and reasoning is carried out by constructing a conditional probability table (CPT); In the continuous implementation, the risk factors are described by a probability density function (PDF), and the joint probability and conditional probability are calculated through integration.

[0075] It should be noted that the discretization method is applicable to scenarios with less data or clear classifications, while the continuousization method is more suitable for scenarios with rich data and the need for precise calculations.

[0076] In one possible implementation, the results of forward reasoning and backward reasoning can be further combined with time series analysis to predict the future trends of risk factors and system states. For example, after identifying foundation settlement as the main risk factor through backward reasoning at a certain time point, the time series data of this factor can be combined to predict its future change trend, thereby predicting the risk evolution process.

[0077] It should be noted that the reasoning analysis by combining time series data can enhance the dynamic adaptability of the model and provide more forward-looking support for subsequent risk control.

[0078] The reasoning analysis process of this embodiment has the following characteristics: Forward reasoning provides a causal prediction path from risk factors to evaluation results, which helps to quantify the overall risk state of the underground system; Backward reasoning provides a reverse interpretation path from evaluation results to risk factors, which helps to identify key risk factors and prioritize control.

[0079] It can be understood that this two-way reasoning mechanism not only enhances the analytical power of the evaluation model, but also improves the reliability and scientificity of risk assessment.

[0080] It should be noted that the forward reasoning and backward reasoning methods disclosed in this embodiment are based on the constructed Bayesian network model and can achieve automated analysis through algorithms.

[0081] It can be understood that the reasoning method in this embodiment is applicable to various underground system scenarios, and the accuracy of its results depends on the accuracy of the conditional probabilities in the Bayesian network and the integrity of the input data. Therefore, in practical applications, the data quality can be improved through a real-time monitoring system, thereby further optimizing the reasoning results.

[0082] For step S5, in this embodiment, by constructing a complex network model of risk factors based on causal relationships, the relationships between various risk factors in the urban underground safety system are further quantified and analyzed. Specifically, the complex network model abstracts risk factors as nodes in the network and causal relationships as edges between nodes, thereby constructing a network model that reflects the topological structure of risk factors and their interactions.

[0083] It should be noted that the purpose of complex network analysis is to identify key risk factors that have an important impact on the overall system risk. The positions and roles of these key factors in the complex network are quantified through network metrics (such as centrality metrics), providing a basis for key control in risk management.

[0084] In some embodiments, the construction of the complex network model is based on the causal relationships in the improved Bayesian network. Specifically, the construction steps include the following: Node definition: Each node in the network represents a risk factor, such as groundwater level, foundation settlement, tunnel deformation, etc.

[0085] Exemplarily, if a certain underground system contains five main risk factors (such as foundation settlement X 1 , groundwater level change X 2 , construction disturbance X 3 , crack development X 4 and surface settlement X 5 ), then the network should contain five corresponding nodes.

[0086] Edge construction: The edges in the network represent the causal relationships between risk factors. As an option, the weight of the edge can be represented by the conditional probability in the Bayesian network, for example: w ij = P(X i |X j ) + P(X j |X i ) It should be noted that this formula represents the comprehensive strength of the two-way causal relationship between node X i and node X j . The higher the weight of the edge, the stronger the correlation between the two risk factors.

[0087] In a possible implementation, the constructed complex network model can be further optimized by a weighted scale-free network model. The characteristic of a weighted scale-free network is that the degree distribution of nodes has a power-law characteristic, that is, a small number of nodes have a large number of connections (referred to as "key nodes"), while most nodes have fewer connections. Through this network characteristic, risk factors that have an important impact on the overall network stability can be identified.

[0088] As an option, in the complex network model, the importance of nodes can be quantified by the following method: Degree centrality: Degree centrality represents the number of direct connections of a certain node and reflects the range directly affected by this node in the network. Its calculation formula is: Where: deg(v i ) represents the number of connections of node v i ; n is the total number of nodes in the network.

[0089] Betweenness centrality: Betweenness centrality represents the control ability of a certain node in the network and calculates the number of times the node appears in all the shortest paths. Its calculation formula is: Where: σ st is the number of shortest paths between nodes s and t; σ st (v i ) represents the number of shortest paths passing through node v i .

[0090] It should be noted that the nodes with higher betweenness centrality are usually key nodes. These nodes play a bridging role in the network and have an important impact on the spread of information or risks.

[0091] In some embodiments, the criticality of risk factors is ranked through a complex network model. For example, for the network model of the above five risk factors, by calculating the degree centrality and betweenness centrality of each node, the following results can be obtained: The betweenness centrality of foundation settlement X 1 is the highest, indicating that it is a key risk factor in the network; The degree centrality of the change in groundwater level X 2 is the highest, indicating that the range it directly affects is the largest.

[0092] It can be understood that such analysis results provide a basis for priority control in the risk management of urban underground systems.

[0093] As a possible implementation method, the construction and analysis of a complex network model can be achieved through a computer program. Specifically: First, extract causal relationship data from the Bayesian network; Secondly, use a graph processing tool (such as the NetworkX library) to construct a complex network; Finally, generate analysis results by calling the calculation function of centrality metrics.

[0094] It should be noted that this implementation method can automatically complete the construction and analysis of a complex network, thereby improving the evaluation efficiency.

[0095] In some embodiments, the analysis results of the complex network model can be visually displayed. For example, use a graphical interface to display the network topology structure, where: The size of the node represents the degree centrality; The thickness of the edge represents the weight of the causal relationship; The key nodes are marked with different colors for highlighting.

[0096] It is understandable that this visualization method can help users quickly understand the distribution and importance of risk factors in the network.

[0097] The complex network model in this embodiment provides a quantitative basis for the priority control of risk factors in the underground system. By ranking the importance of nodes, it can help managers identify key risk factors and concentrate resources on monitoring and managing them.

[0098] For step S6, in this embodiment, based on the theory of stochastic dynamic systems, the temporal evolution of risk factors is modeled, and the optimization of urban underground safety risk assessment is achieved through the dynamic update of real-time data. Specifically, by establishing a dynamic evolution equation for the state of risk factors to describe its trend of change over time, and combining real-time monitoring data to adjust the model parameters and evaluation results, so as to ensure that the risk assessment can accurately reflect the current state of the system.

[0099] It should be noted that the modeling of the temporal evolution of risk factors can not only capture the change law in the time dimension, but also improve the adaptability of the model through dynamic update, providing support for subsequent evaluation and management.

[0100] In some embodiments, the temporal evolution model of risk factors can be described by a stochastic dynamics equation in the following form: X(t + 1) = X(t) + Δt·f(X(t)) + ∈ Where: X(t) represents the state of the risk factor at time t; Δt is the time step; f(X(t)) is the evolution function of the risk factor, representing its trend of change over time; ∈ is a random perturbation term, indicating the influence of external uncertain factors on the risk factor.

[0101] As an option, the form of the evolution function f(X(t)) can be designed according to the requirements of the actual scenario. For example, for the change of the groundwater level, its change rate can be associated with factors such as groundwater seepage flow and precipitation to establish corresponding linear or nonlinear relationships.

[0102] Specifically, the introduction of the random perturbation term ∈ is to simulate the randomness and uncertainty in the evolution of risk factors. Exemplarily, the change of the groundwater level may be affected by uncontrollable external factors (such as sudden rainfall or surface seepage), causing its state to deviate from the normal change trend.

[0103] In a possible implementation, the random perturbation term ∈ can be assumed to follow a Gaussian distribution, and its mathematical expression is: Wherein: represents a normal distribution with a mean of 0 and a variance of σ 2 ; σ 2 represents the intensity of random perturbation and is used to control the amplitude of randomness.

[0104] It can be understood that by adjusting the parameters of the random perturbation term, the influence of different degrees of uncertainty on the risk factor can be simulated.

[0105] In some embodiments, in order to update the dynamic changes of the risk factor in real time, the present embodiment corrects the state X(t) and model parameters in combination with real-time monitoring data. Specifically, the following update formula can be adopted: X′(t) = α·X real (t) + (1 - α)·X(t) Wherein: X′(t) represents the corrected risk factor state; X real (t) represents the actual state of the risk factor obtained through real-time monitoring; α is an update weight parameter and is used to balance the influence between the monitoring data and the model prediction value.

[0106] It should be noted that the value of the update weight α can be dynamically adjusted according to the accuracy and reliability of the monitoring data. For example, when the noise of the monitoring data is large, the value of α can be reduced to reduce the dependence on the monitoring data.

[0107] In a possible implementation manner, in order to enhance the prediction ability of the model for long-term trends, the present embodiment also introduces a time series analysis method. For example, the long-term change trend of the risk factor state can be modeled in combination with historical data and combined with the short-term prediction results of the stochastic dynamic system to generate a more comprehensive risk factor state assessment.

[0108] Exemplarily, the time series analysis can adopt the following form: Wherein: X(t + k) represents the predicted state at time t + k; β i is the weight parameter of the time series model; m is the length of the historical data used; γ is a trend correction term.

[0109] It should be noted that the combination of time series analysis and the stochastic dynamic system can more accurately capture the dynamic change law of the risk factor and improve the prediction ability of the model.

[0110] In some embodiments, the results of real-time update and dynamic adjustment can be presented in a visual manner. For example, a time-series curve of the risk factor status can be plotted, where: The solid line represents the state change predicted by the model; The dashed line represents the change in real-time monitoring data; The shaded area represents the scope of influence of random perturbations.

[0111] It can be understood that this visual manner can intuitively reflect the dynamic changes of risk factors and the adaptability of the model.

[0112] The time-series evolution modeling and dynamic update method in this embodiment can effectively solve the time dimension problem in urban underground safety assessment. By combining the theory of stochastic dynamic systems and real-time data, the present invention can maintain the accuracy and real-time performance of the assessment model in a complex and changeable environment.

[0113] For step S7, in this embodiment, by comprehensively processing the data and model calculation results generated in the foregoing steps, the assessment result of the urban underground safety risk is finally output. Specifically, the output assessment result includes the overall risk status of the underground system, the list of key risk factors, and the prediction of the risk change trend. These results are presented in a quantitative form, providing a scientific basis for risk management and decision-making.

[0114] It should be noted that the output of the assessment result can be presented in combination with the calculation result of the model and a visualization tool, so as to intuitively reflect the safety risk status of the underground system.

[0115] In some embodiments, the overall risk status of the underground system is an important part of the assessment result. The overall risk status is generated by normalizing the comprehensive scores of all risk factors. The specific calculation formula is as follows: Where: R total Represents the score of the overall risk status; w i Represents the weight of the i-th risk factor, which can be obtained from the weight optimization in step S2 above; R i Represents the score of the i-th risk factor, which can be directly obtained from the risk assessment matrix; n represents the total number of risk factors.

[0116] As a possible implementation, the score of the overall risk status can be divided into several levels. For example: Low risk (score less than 3): The underground system operates stably and no additional measures are required; Medium risk (score between 3 and 6): It is recommended to strengthen monitoring and pay attention to potential hazards; High risk (score greater than 6): Immediate risk control measures need to be taken to avoid serious consequences.

[0117] It can be understood that this quantified risk status provides an intuitive evaluation of the system security status for managers.

[0118] In this embodiment, the output of the list of key risk factors is based on the analysis results of the complex network model in the foregoing step S5. Specifically, by analyzing the centrality indicators of network nodes (such as degree centrality and betweenness centrality), the risk factors that have the greatest impact on the overall system risk can be determined.

[0119] Exemplarily, for the complex network model of a certain underground system, the key risk factors may include the following: Foundation settlement (with the highest betweenness centrality), indicating that it plays a bridging role in the risk propagation path; Change in groundwater level (with the highest degree centrality), indicating that the risk range it directly affects is the largest.

[0120] As an option, the key risk factors can be sorted according to importance and their specific contribution values to the overall risk can be marked. The calculation formula is: Where: C impact (X i ) represents the relative importance of risk factor X i ; deg(X i ) represents the number of connections of node X i .

[0121] It should be noted that the output of the key risk factors not only helps to identify the weak links of the system, but also provides a scientific basis for priority control and resource allocation.

[0122] In some embodiments, the risk change trend prediction is generated by combining a stochastic dynamic system model and time series analysis. Specifically, the dynamic evolution of the risk factors in the foregoing step S6 provides data support for trend prediction. The calculation formula for trend prediction is as follows: Where: X(t + k) represents the predicted state of the risk factor after k time steps in the future; X(t - i) represents the state of the risk factor at the i-th time step in the past; β iis the weight parameter of the time series model; ∈ is the random disturbance term.

[0123] As an option, the prediction result can be presented in the form of a trend curve, for example: The ascending section of the curve indicates that the state of the risk factor is deteriorating; The flat section of the curve indicates that the state of the risk factor remains stable.

[0124] It should be noted that the result of trend prediction can help managers identify potential changes in risks in advance, so as to formulate more effective risk control plans.

[0125] In a possible implementation, the evaluation result can be presented through a data visualization tool. For example, the evaluation result is displayed in the following ways: The overall risk status is displayed in the form of a dashboard or a bar chart, intuitively reflecting the safety level of the underground system; The key risk factors are displayed in the form of a list or a radar chart, and their centrality indicators are attached; The risk change trend is displayed in the form of a line chart, dynamically showing the time evolution process of the risk factor.

[0126] It can be understood that this visualization display method can effectively improve the readability of the evaluation result and facilitate managers to quickly understand the risk status of the system.

[0127] The evaluation result generation and output method in this embodiment can effectively integrate the analysis results of the foregoing steps, providing comprehensive support for the evaluation and management of urban underground safety risks. By outputting the overall risk status, key risk factors and risk change trends, managers can form a global understanding of the safety status of the underground system and take targeted risk control measures.

[0128] Generally speaking, the present invention realizes the scientific quantification, real-time update and dynamic evaluation of urban underground safety risks by constructing a risk assessment matrix, optimizing the weights of risk factors, causal reasoning analysis based on the Bayesian network model, identification of key factors of the complex network model, and dynamic time series modeling combined with the stochastic dynamic system. The method includes using the entropy value theory to adjust the weights to reflect the actual contributions of risk factors, conducting forward and backward reasoning analysis based on the improved Bayesian network model, and combining complex network theory to identify key risk factors, and finally outputting the overall risk status, key factors and risk change trends. Through a systematic and data-driven method, the invention provides an efficient, accurate and dynamically adaptable technical solution for urban underground safety management.

[0129] The urban underground safety risk assessment device based on the dynamic update algorithm described below can be correspondingly referred to the urban underground safety risk assessment method based on the dynamic update algorithm described above.

[0130] Please refer to the appendix Figure 2 , the present invention also provides an urban underground safety risk assessment device based on the dynamic update algorithm, including: a matrix construction module 100, configured to quantify the consequence severity level and occurrence probability level of risk factors and construct a risk assessment matrix; a weight optimization module 200, configured to dynamically adjust the weights of risk factors based on the entropy value theory; a network construction module 300, configured to construct an improved Bayesian network based on risk factors and their causal relationships; an inference analysis module 400, configured to perform forward inference and backward inference based on the Bayesian network model; a network analysis module 500, configured to construct a complex network based on causal relationships and identify key risk factors; a dynamic update module 600, configured to model and update the temporal evolution of risk factors in real time.

[0131] The device of this embodiment can be used to execute the above method embodiment, and its principle and technical effect are similar, which will not be elaborated here.

[0132] Please refer to the appendix Figure 3 , the present invention also provides an electronic device 40, including: a processor 41 and a memory 42, the memory 42 stores a computer program executable by the processor, and when the computer program is executed by the processor, it executes the above method.

[0133] The present invention also provides a storage medium 43, on which a computer program is stored, and when the computer program is run by the processor 41, it executes the above method.

[0134] Among them, the storage medium 43 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0135] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An urban underground safety risk assessment method based on a dynamic update algorithm, characterized in that: The following steps are involved: Construct an urban underground safety risk assessment matrix to quantify the risk factors of urban underground safety based on the severity level and probability level of the risk factors; Optimize the weights of risk factors, adopt entropy theory, and dynamically adjust the weights according to the contribution of risk factors to the assessment results; Based on risk factors and their causal relationships, an improved Bayesian network model is constructed, in which risk factors are used as network nodes and causal relationships are used as network edges. The network structure is optimized by combining a dynamic causal relationship adjustment algorithm. The improved Bayesian network model is used for forward reasoning and reverse reasoning to predict the safety risk status of the underground system and identify key risk factors. Construct a complex network model of risk factors based on causal relationships, analyze the importance of network nodes, and identify key risk factors; Based on the theory of random dynamic systems, the temporal evolution of risk factors is modeled, and the dynamic changes of risk factors are updated and adjusted in real time; Output assessment results, including the overall risk status of the underground system, key risk factors and risk change trends.

2. The urban underground safety risk assessment method based on the dynamic update algorithm according to claim 1 is characterized in that: The steps of constructing the urban underground safety risk assessment matrix include: By dividing the severity level of the consequences of risk factors into multiple levels and the probability level of occurrence into multiple levels, a two-dimensional matrix is ​​generated, in which the comprehensive score of the risk factor is the product of the severity level of the consequences and the probability level of occurrence.

3. The urban underground safety risk assessment method based on the dynamic update algorithm according to claim 1 is characterized in that: The step of optimizing the weights of risk factors comprises: Based on the entropy theory, the conditional entropy of each risk factor and the assessment result is calculated, and the weight of the risk factor is normalized and adjusted according to the conditional entropy value, so as to dynamically optimize the impact weight of the risk factor.

4. The urban underground safety risk assessment method based on the dynamic update algorithm according to claim 1 is characterized in that: The steps of constructing an improved Bayesian network model based on risk factors and their causal relationships include: Through the Markov Chain Monte Carlo algorithm, the causal relationship edge weights between risk factors in the network are iteratively adjusted to optimize the causal edge weights, and a dynamic Bayesian network model is constructed based on the update of the causal relationship.

5. The urban underground safety risk assessment method based on the dynamic update algorithm according to claim 1 is characterized in that: The steps of using the improved Bayesian network model to perform forward reasoning and reverse reasoning include: Calculate the probability of occurrence of potential risk consequences of the underground system based on the current state of risk factors; Based on the risk consequences of the underground system, reversely deduce the main risk factors that cause the risk.

6. The urban underground safety risk assessment method based on the dynamic update algorithm according to claim 1 is characterized in that: The steps of constructing a risk factor complex network model based on causal relationships include: Risk factors are regarded as network nodes, and the causal relationships between risk factors are regarded as network edges. A weighted scale-free network is constructed according to the weights of the causal relationships, and key risk factors are identified by analyzing the degree centrality and betweenness centrality of network nodes.

7. The urban underground safety risk assessment method based on the dynamic update algorithm according to claim 1 is characterized in that: The steps of modeling the temporal evolution of risk factors based on the theory of random dynamic systems and updating and adjusting the dynamic changes of risk factors in real time include: By establishing a dynamic evolution model of the risk factor status, describing the changing trend of the risk factors, and adjusting the state probability and edge weights of the nodes in the Bayesian network in combination with real-time data, the real-time update of the evaluation model can be achieved.

8. An urban underground safety risk assessment device based on a dynamic update algorithm, used to execute the method according to any one of claims 1 to 7, characterized in that: include: The matrix construction module is used to quantify the severity level and probability level of the risk factors and construct a risk assessment matrix; Weight optimization module, used to dynamically adjust the weights of risk factors based on entropy theory; A network building module, used to construct an improved Bayesian network based on risk factors and their causal relationships; Reasoning analysis module, used for forward reasoning and reverse reasoning based on Bayesian network model; Network analysis module, used to construct complex networks based on causal relationships and identify key risk factors; Dynamic update module, used to model and update the temporal evolution of risk factors in real time.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images